Compare commits
2 Commits
43ad813e0d
...
feat/feder
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
8bd288a7db | ||
|
|
8df6ef4c51 |
@@ -4,7 +4,6 @@ import { dirname, join, resolve } from 'node:path';
|
|||||||
import { Command } from 'commander';
|
import { Command } from 'commander';
|
||||||
import { afterEach, describe, expect, it, vi } from 'vitest';
|
import { afterEach, describe, expect, it, vi } from 'vitest';
|
||||||
import {
|
import {
|
||||||
acquireRestartLock,
|
|
||||||
addAgentToRoster,
|
addAgentToRoster,
|
||||||
buildAgentSendCommand,
|
buildAgentSendCommand,
|
||||||
buildAgentWatchAttachCommand,
|
buildAgentWatchAttachCommand,
|
||||||
@@ -46,8 +45,6 @@ import {
|
|||||||
removeAgentFromRoster,
|
removeAgentFromRoster,
|
||||||
resolveFleetPaths,
|
resolveFleetPaths,
|
||||||
resolvePresetFilename,
|
resolvePresetFilename,
|
||||||
restartLockPath,
|
|
||||||
RESTART_LOCK_STALE_MS,
|
|
||||||
RUNTIME_ACCEPTABLE_COMMANDS,
|
RUNTIME_ACCEPTABLE_COMMANDS,
|
||||||
serializeRosterToYaml,
|
serializeRosterToYaml,
|
||||||
VERIFY_DEFAULT_TIMEOUT_MS,
|
VERIFY_DEFAULT_TIMEOUT_MS,
|
||||||
@@ -681,231 +678,6 @@ describe('fleet command construction', () => {
|
|||||||
}
|
}
|
||||||
});
|
});
|
||||||
|
|
||||||
it('waits for an in-flight restart to clear before relaunching (re-entry guard)', async () => {
|
|
||||||
const home = await tempDir();
|
|
||||||
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
|
||||||
await mkdir(join(home, 'fleet'), { recursive: true });
|
|
||||||
await writeFile(
|
|
||||||
rosterPath,
|
|
||||||
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
|
|
||||||
'\n',
|
|
||||||
),
|
|
||||||
);
|
|
||||||
|
|
||||||
// Simulate another `mosaic fleet restart` process mid-teardown: a fresh lock
|
|
||||||
// (recent timestamp, so it is NOT treated as stale) already held.
|
|
||||||
const lockPath = restartLockPath(home);
|
|
||||||
await mkdir(dirname(lockPath), { recursive: true });
|
|
||||||
await writeFile(lockPath, `4242\n${Date.now()}\n`);
|
|
||||||
|
|
||||||
const events: string[] = [];
|
|
||||||
const runner: CommandRunner = async (command, args) => {
|
|
||||||
events.push(`run:${args[args.length - 1]}`);
|
|
||||||
return { stdout: '', stderr: '', exitCode: 0 };
|
|
||||||
};
|
|
||||||
// The injected sleep stands in for time passing while we wait; the in-flight
|
|
||||||
// restart "finishes" (releases its lock) after the first poll.
|
|
||||||
let sleeps = 0;
|
|
||||||
const sleepFn: SleepFn = async () => {
|
|
||||||
sleeps += 1;
|
|
||||||
events.push(`sleep:${sleeps}`);
|
|
||||||
await rm(lockPath, { force: true });
|
|
||||||
};
|
|
||||||
|
|
||||||
const program = new Command();
|
|
||||||
program.exitOverride();
|
|
||||||
registerFleetCommand(program, { runner, sleepFn, mosaicHome: home });
|
|
||||||
|
|
||||||
try {
|
|
||||||
await program.parseAsync(['node', 'mosaic', 'fleet', 'restart']);
|
|
||||||
|
|
||||||
// It must have waited at least once before issuing any systemctl restart.
|
|
||||||
expect(sleeps).toBeGreaterThan(0);
|
|
||||||
const firstSleep = events.findIndex((e) => e.startsWith('sleep:'));
|
|
||||||
const firstRun = events.findIndex((e) => e.startsWith('run:'));
|
|
||||||
expect(firstSleep).toBeGreaterThanOrEqual(0);
|
|
||||||
expect(firstRun).toBeGreaterThan(firstSleep);
|
|
||||||
|
|
||||||
// And it still performs the full restart once the lock clears.
|
|
||||||
expect(events).toContain('run:mosaic-tmux-holder.service');
|
|
||||||
expect(events).toContain('run:mosaic-agent@coder0.service');
|
|
||||||
|
|
||||||
// The lock is released after the restart completes.
|
|
||||||
await expect(readFile(lockPath, 'utf8')).rejects.toMatchObject({ code: 'ENOENT' });
|
|
||||||
} finally {
|
|
||||||
await rm(home, { recursive: true, force: true });
|
|
||||||
}
|
|
||||||
});
|
|
||||||
|
|
||||||
it('breaks a stale restart lock and proceeds without waiting', async () => {
|
|
||||||
const home = await tempDir();
|
|
||||||
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
|
||||||
await mkdir(join(home, 'fleet'), { recursive: true });
|
|
||||||
await writeFile(
|
|
||||||
rosterPath,
|
|
||||||
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
|
|
||||||
'\n',
|
|
||||||
),
|
|
||||||
);
|
|
||||||
|
|
||||||
// A lock left behind by a crashed owner: timestamp older than the stale window.
|
|
||||||
const lockPath = restartLockPath(home);
|
|
||||||
await mkdir(dirname(lockPath), { recursive: true });
|
|
||||||
await writeFile(lockPath, `4242\n${Date.now() - RESTART_LOCK_STALE_MS - 1_000}\n`);
|
|
||||||
|
|
||||||
const calls: string[][] = [];
|
|
||||||
const runner: CommandRunner = async (command, args) => {
|
|
||||||
calls.push([command, ...args]);
|
|
||||||
return { stdout: '', stderr: '', exitCode: 0 };
|
|
||||||
};
|
|
||||||
const sleepFn = vi.fn<SleepFn>(async () => {});
|
|
||||||
|
|
||||||
const program = new Command();
|
|
||||||
program.exitOverride();
|
|
||||||
registerFleetCommand(program, { runner, sleepFn, mosaicHome: home });
|
|
||||||
|
|
||||||
try {
|
|
||||||
await program.parseAsync(['node', 'mosaic', 'fleet', 'restart']);
|
|
||||||
|
|
||||||
// Stale lock is broken immediately — no waiting.
|
|
||||||
expect(sleepFn).not.toHaveBeenCalled();
|
|
||||||
expect(calls).toEqual([
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-tmux-holder.service'],
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-agent@coder0.service'],
|
|
||||||
]);
|
|
||||||
// The stale lock is gone once the restart completes.
|
|
||||||
await expect(readFile(lockPath, 'utf8')).rejects.toMatchObject({ code: 'ENOENT' });
|
|
||||||
} finally {
|
|
||||||
await rm(home, { recursive: true, force: true });
|
|
||||||
}
|
|
||||||
});
|
|
||||||
|
|
||||||
it('releases the restart lock so a subsequent restart is not blocked', async () => {
|
|
||||||
const home = await tempDir();
|
|
||||||
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
|
||||||
await mkdir(join(home, 'fleet'), { recursive: true });
|
|
||||||
await writeFile(
|
|
||||||
rosterPath,
|
|
||||||
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
|
|
||||||
'\n',
|
|
||||||
),
|
|
||||||
);
|
|
||||||
|
|
||||||
const calls: string[][] = [];
|
|
||||||
const runner: CommandRunner = async (command, args) => {
|
|
||||||
calls.push([command, ...args]);
|
|
||||||
return { stdout: '', stderr: '', exitCode: 0 };
|
|
||||||
};
|
|
||||||
const sleepFn = vi.fn<SleepFn>(async () => {});
|
|
||||||
|
|
||||||
const program = new Command();
|
|
||||||
program.exitOverride();
|
|
||||||
registerFleetCommand(program, { runner, sleepFn, mosaicHome: home });
|
|
||||||
|
|
||||||
try {
|
|
||||||
await program.parseAsync(['node', 'mosaic', 'fleet', 'restart']);
|
|
||||||
await program.parseAsync(['node', 'mosaic', 'fleet', 'restart']);
|
|
||||||
|
|
||||||
// Two sequential restarts both run fully and neither has to wait.
|
|
||||||
expect(sleepFn).not.toHaveBeenCalled();
|
|
||||||
expect(calls).toEqual([
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-tmux-holder.service'],
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-agent@coder0.service'],
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-tmux-holder.service'],
|
|
||||||
['systemctl', '--user', 'restart', 'mosaic-agent@coder0.service'],
|
|
||||||
]);
|
|
||||||
} finally {
|
|
||||||
await rm(home, { recursive: true, force: true });
|
|
||||||
}
|
|
||||||
});
|
|
||||||
|
|
||||||
it('guards the single-agent restart path behind the in-flight restart lock', async () => {
|
|
||||||
const home = await tempDir();
|
|
||||||
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
|
||||||
await mkdir(join(home, 'fleet'), { recursive: true });
|
|
||||||
await writeFile(
|
|
||||||
rosterPath,
|
|
||||||
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
|
|
||||||
'\n',
|
|
||||||
),
|
|
||||||
);
|
|
||||||
|
|
||||||
// A full restart is mid-flight (lock held); a single-agent restart re-enters.
|
|
||||||
const lockPath = restartLockPath(home);
|
|
||||||
await mkdir(dirname(lockPath), { recursive: true });
|
|
||||||
await writeFile(lockPath, `4242\n${Date.now()}\n`);
|
|
||||||
|
|
||||||
const events: string[] = [];
|
|
||||||
const runner: CommandRunner = async (command, args) => {
|
|
||||||
events.push(`run:${args[args.length - 1]}`);
|
|
||||||
return { stdout: '', stderr: '', exitCode: 0 };
|
|
||||||
};
|
|
||||||
let sleeps = 0;
|
|
||||||
const sleepFn: SleepFn = async () => {
|
|
||||||
sleeps += 1;
|
|
||||||
events.push(`sleep:${sleeps}`);
|
|
||||||
await rm(lockPath, { force: true });
|
|
||||||
};
|
|
||||||
|
|
||||||
const program = new Command();
|
|
||||||
program.exitOverride();
|
|
||||||
registerFleetCommand(program, { runner, sleepFn, mosaicHome: home });
|
|
||||||
|
|
||||||
try {
|
|
||||||
await program.parseAsync(['node', 'mosaic', 'fleet', 'restart', 'coder0']);
|
|
||||||
|
|
||||||
// The single-agent restart waits for the in-flight restart before acting.
|
|
||||||
expect(sleeps).toBeGreaterThan(0);
|
|
||||||
const firstSleep = events.findIndex((e) => e.startsWith('sleep:'));
|
|
||||||
const firstRun = events.findIndex((e) => e.startsWith('run:'));
|
|
||||||
expect(firstSleep).toBeGreaterThanOrEqual(0);
|
|
||||||
expect(firstRun).toBeGreaterThan(firstSleep);
|
|
||||||
// Only the named agent is restarted; the holder is untouched.
|
|
||||||
expect(events).toContain('run:mosaic-agent@coder0.service');
|
|
||||||
expect(events).not.toContain('run:mosaic-tmux-holder.service');
|
|
||||||
} finally {
|
|
||||||
await rm(home, { recursive: true, force: true });
|
|
||||||
}
|
|
||||||
});
|
|
||||||
|
|
||||||
it('does not let a timed-out owner drop a lock another restart broke and re-owned', async () => {
|
|
||||||
const home = await tempDir();
|
|
||||||
const runDir = join(home, 'fleet', 'run');
|
|
||||||
await mkdir(runDir, { recursive: true });
|
|
||||||
const lockPath = restartLockPath(home);
|
|
||||||
const tokenOf = async (): Promise<string> => {
|
|
||||||
const raw = await readFile(lockPath, 'utf8');
|
|
||||||
return raw.split('\n')[2]?.trim() ?? '';
|
|
||||||
};
|
|
||||||
const sleepFn = vi.fn<SleepFn>(async () => {});
|
|
||||||
|
|
||||||
// R1 acquires the lock and begins a restart that then hangs.
|
|
||||||
const r1 = await acquireRestartLock(home, sleepFn);
|
|
||||||
const tokenR1 = await tokenOf();
|
|
||||||
expect(tokenR1).not.toBe('');
|
|
||||||
|
|
||||||
// The hung R1 leaves a stale lock: rewrite its timestamp into the past while
|
|
||||||
// preserving R1's token — exactly the on-disk state a stuck owner leaves.
|
|
||||||
await writeFile(lockPath, `4242\n${Date.now() - RESTART_LOCK_STALE_MS - 1_000}\n${tokenR1}\n`);
|
|
||||||
|
|
||||||
// R2 re-enters, sees the stale lock, and atomically takes ownership.
|
|
||||||
const r2 = await acquireRestartLock(home, sleepFn);
|
|
||||||
const tokenR2 = await tokenOf();
|
|
||||||
expect(tokenR2).not.toBe(tokenR1);
|
|
||||||
expect(sleepFn).not.toHaveBeenCalled();
|
|
||||||
|
|
||||||
// R1 finally finishes and releases. It must NOT delete R2's lock — otherwise
|
|
||||||
// a third restart (R3) could acquire and interleave with R2 still running.
|
|
||||||
await r1.release();
|
|
||||||
expect(await tokenOf()).toBe(tokenR2);
|
|
||||||
|
|
||||||
// R2 releases cleanly and the lock is gone.
|
|
||||||
await r2.release();
|
|
||||||
await expect(readFile(lockPath, 'utf8')).rejects.toMatchObject({ code: 'ENOENT' });
|
|
||||||
|
|
||||||
await rm(home, { recursive: true, force: true });
|
|
||||||
});
|
|
||||||
|
|
||||||
it('attempts every agent and the holder during fleet stop even when an agent stop fails', async () => {
|
it('attempts every agent and the holder during fleet stop even when an agent stop fails', async () => {
|
||||||
const home = await tempDir();
|
const home = await tempDir();
|
||||||
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
const rosterPath = join(home, 'fleet', 'roster.yaml');
|
||||||
|
|||||||
@@ -1,16 +1,5 @@
|
|||||||
import { constants } from 'node:fs';
|
import { constants } from 'node:fs';
|
||||||
import {
|
import { access, chmod, copyFile, mkdir, readFile, unlink, writeFile } from 'node:fs/promises';
|
||||||
access,
|
|
||||||
chmod,
|
|
||||||
copyFile,
|
|
||||||
mkdir,
|
|
||||||
open,
|
|
||||||
readFile,
|
|
||||||
rename,
|
|
||||||
unlink,
|
|
||||||
writeFile,
|
|
||||||
} from 'node:fs/promises';
|
|
||||||
import { randomUUID } from 'node:crypto';
|
|
||||||
import { homedir, hostname, userInfo } from 'node:os';
|
import { homedir, hostname, userInfo } from 'node:os';
|
||||||
import { dirname, join, resolve } from 'node:path';
|
import { dirname, join, resolve } from 'node:path';
|
||||||
import { fileURLToPath } from 'node:url';
|
import { fileURLToPath } from 'node:url';
|
||||||
@@ -544,173 +533,6 @@ export function buildFleetServiceCommand(action: FleetServiceAction, agentName?:
|
|||||||
return ['systemctl', '--user', action, service];
|
return ['systemctl', '--user', action, service];
|
||||||
}
|
}
|
||||||
|
|
||||||
/** Poll interval (ms) while waiting for an in-flight restart's lock to clear. */
|
|
||||||
export const RESTART_LOCK_POLL_INTERVAL_MS = 250;
|
|
||||||
/**
|
|
||||||
* Maximum time (ms) a re-entrant restart waits for the in-flight restart to
|
|
||||||
* finish before it breaks the lock and proceeds anyway. A bound is required so
|
|
||||||
* a crashed holder of the lock can never deadlock the fleet permanently.
|
|
||||||
*/
|
|
||||||
export const RESTART_LOCK_MAX_WAIT_MS = 30_000;
|
|
||||||
/**
|
|
||||||
* Age (ms) past which a restart lock is treated as stale (its owner died
|
|
||||||
* without releasing it) and is broken immediately rather than waited on.
|
|
||||||
*/
|
|
||||||
export const RESTART_LOCK_STALE_MS = 60_000;
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Resolves the path of the cross-process restart lock for a given Mosaic home.
|
|
||||||
* Kept strictly under `<mosaicHome>/fleet/run` (not the heartbeat env override)
|
|
||||||
* so the lock is scoped to the same fleet the restart acts on.
|
|
||||||
*/
|
|
||||||
export function restartLockPath(mosaicHome: string): string {
|
|
||||||
return join(mosaicHome, 'fleet', 'run', 'restart.lock');
|
|
||||||
}
|
|
||||||
|
|
||||||
/** A held restart lock; `release()` removes the lock file iff we still own it. */
|
|
||||||
interface RestartGuard {
|
|
||||||
release(): Promise<void>;
|
|
||||||
}
|
|
||||||
|
|
||||||
/** Lock-file contents: pid (informational), timestamp, and a unique owner token. */
|
|
||||||
function formatRestartLockContent(token: string): string {
|
|
||||||
return `${process.pid}\n${Date.now()}\n${token}\n`;
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Reads the owner token (line 3) from a lock file, or null if the file is
|
|
||||||
* missing/unreadable/tokenless. The token is what makes release and break
|
|
||||||
* ownership-safe: a process only ever acts on a lock whose token matches its own.
|
|
||||||
*/
|
|
||||||
async function readRestartLockToken(lockPath: string): Promise<string | null> {
|
|
||||||
let raw: string;
|
|
||||||
try {
|
|
||||||
raw = await readFile(lockPath, 'utf8');
|
|
||||||
} catch {
|
|
||||||
return null;
|
|
||||||
}
|
|
||||||
const token = raw.split('\n')[2]?.trim();
|
|
||||||
return token ? token : null;
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Returns true when an existing lock file is stale: older than
|
|
||||||
* RESTART_LOCK_STALE_MS, or unreadable/unparseable (a corrupt or partially
|
|
||||||
* written lock left by a crashed owner). A vanished lock (ENOENT) is not stale —
|
|
||||||
* the next acquire attempt will simply succeed.
|
|
||||||
*/
|
|
||||||
async function isRestartLockStale(lockPath: string, now: number): Promise<boolean> {
|
|
||||||
let raw: string;
|
|
||||||
try {
|
|
||||||
raw = await readFile(lockPath, 'utf8');
|
|
||||||
} catch (err) {
|
|
||||||
if ((err as NodeJS.ErrnoException).code === 'ENOENT') {
|
|
||||||
return false;
|
|
||||||
}
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
const stampLine = raw.split('\n')[1] ?? '';
|
|
||||||
const stamp = Number.parseInt(stampLine.trim(), 10);
|
|
||||||
if (!Number.isFinite(stamp)) {
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
return now - stamp >= RESTART_LOCK_STALE_MS;
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Atomically take over an existing (stale or timed-out) lock WITHOUT blind
|
|
||||||
* unlinking it: write our own token to a temp file and `rename()` it over the
|
|
||||||
* lock. rename is atomic, so it replaces the prior owner's content in one step
|
|
||||||
* rather than the unsafe unlink-then-recreate (which a third restart could slip
|
|
||||||
* between). Returns true only if our token is the one on disk afterwards — if a
|
|
||||||
* concurrent breaker raced and won, we read back their token and return false so
|
|
||||||
* the caller keeps waiting instead of assuming ownership.
|
|
||||||
*/
|
|
||||||
async function breakAndOwnRestartLock(
|
|
||||||
lockPath: string,
|
|
||||||
token: string,
|
|
||||||
content: string,
|
|
||||||
): Promise<boolean> {
|
|
||||||
const tmpPath = `${lockPath}.${token}`;
|
|
||||||
await writeFile(tmpPath, content);
|
|
||||||
try {
|
|
||||||
await rename(tmpPath, lockPath);
|
|
||||||
} catch (err) {
|
|
||||||
await unlink(tmpPath).catch(() => {});
|
|
||||||
throw err;
|
|
||||||
}
|
|
||||||
return (await readRestartLockToken(lockPath)) === token;
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Acquire the fleet restart lock, serializing concurrent `mosaic fleet restart`
|
|
||||||
* invocations across processes. Each restart tears the tmux holder (and the
|
|
||||||
* agent sessions inside it) down and back up; without this guard a re-entrant
|
|
||||||
* restart relaunches agents against a half-torn-down holder, which fails and
|
|
||||||
* tight-loops. A re-entrant caller waits for the in-flight restart to release
|
|
||||||
* the lock (clean shutdown settled) before proceeding, breaks a stale lock left
|
|
||||||
* by a crashed owner, and after RESTART_LOCK_MAX_WAIT_MS breaks the lock to
|
|
||||||
* avoid a permanent deadlock.
|
|
||||||
*
|
|
||||||
* Ownership is tracked by a unique per-acquire token written into the lock.
|
|
||||||
* `release()` only unlinks the lock while our token is still on disk, and a
|
|
||||||
* break takes ownership atomically — so once another caller has broken and
|
|
||||||
* re-owned the lock, neither the timed-out original owner's `release()` nor a
|
|
||||||
* stale `break` can drop the new owner's lock and let a third restart interleave.
|
|
||||||
*/
|
|
||||||
export async function acquireRestartLock(
|
|
||||||
mosaicHome: string,
|
|
||||||
sleepFn: SleepFn,
|
|
||||||
): Promise<RestartGuard> {
|
|
||||||
const token = randomUUID();
|
|
||||||
const lockPath = restartLockPath(mosaicHome);
|
|
||||||
await mkdir(dirname(lockPath), { recursive: true });
|
|
||||||
const release = async (): Promise<void> => {
|
|
||||||
// Ownership-safe: only remove the lock if it is still ours. If another
|
|
||||||
// caller broke and re-owned it (after a stale/timeout break), the token no
|
|
||||||
// longer matches and we must leave their lock intact.
|
|
||||||
if ((await readRestartLockToken(lockPath)) !== token) {
|
|
||||||
return;
|
|
||||||
}
|
|
||||||
try {
|
|
||||||
await unlink(lockPath);
|
|
||||||
} catch {
|
|
||||||
// Raced away between the token check and unlink — nothing more to do.
|
|
||||||
}
|
|
||||||
};
|
|
||||||
const deadline = Date.now() + RESTART_LOCK_MAX_WAIT_MS;
|
|
||||||
for (;;) {
|
|
||||||
try {
|
|
||||||
const handle = await open(lockPath, 'wx');
|
|
||||||
await handle.writeFile(formatRestartLockContent(token));
|
|
||||||
await handle.close();
|
|
||||||
return { release };
|
|
||||||
} catch (err) {
|
|
||||||
if ((err as NodeJS.ErrnoException).code !== 'EEXIST') {
|
|
||||||
throw err;
|
|
||||||
}
|
|
||||||
// A restart is already in flight (or its lock was left behind).
|
|
||||||
const stale = await isRestartLockStale(lockPath, Date.now());
|
|
||||||
const timedOut = Date.now() >= deadline;
|
|
||||||
if (stale || timedOut) {
|
|
||||||
if (await breakAndOwnRestartLock(lockPath, token, formatRestartLockContent(token))) {
|
|
||||||
process.stderr.write(
|
|
||||||
stale
|
|
||||||
? 'Breaking stale fleet restart lock and proceeding.\n'
|
|
||||||
: `Timed out after ${RESTART_LOCK_MAX_WAIT_MS}ms waiting for the in-flight fleet ` +
|
|
||||||
'restart; breaking the lock and proceeding.\n',
|
|
||||||
);
|
|
||||||
return { release };
|
|
||||||
}
|
|
||||||
// A concurrent breaker won the takeover; back off and re-evaluate.
|
|
||||||
await sleepFn(RESTART_LOCK_POLL_INTERVAL_MS);
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
await sleepFn(RESTART_LOCK_POLL_INTERVAL_MS);
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* Returns the systemctl --user enable command for a given unit.
|
* Returns the systemctl --user enable command for a given unit.
|
||||||
* Used by the install auto-enable step to persist units across reboots.
|
* Used by the install auto-enable step to persist units across reboots.
|
||||||
@@ -1350,7 +1172,6 @@ export function isSendAccepted(capturedOutput: string): SendVerifyResult {
|
|||||||
|
|
||||||
export function registerFleetCommand(program: Command, deps: FleetCommandDeps = {}): Command {
|
export function registerFleetCommand(program: Command, deps: FleetCommandDeps = {}): Command {
|
||||||
const runner = deps.runner ?? runCommand;
|
const runner = deps.runner ?? runCommand;
|
||||||
const sleepFn = deps.sleepFn ?? defaultSleep;
|
|
||||||
const paths = resolveFleetPaths(deps.mosaicHome);
|
const paths = resolveFleetPaths(deps.mosaicHome);
|
||||||
const frameworkRoot = deps.frameworkRoot ?? resolveFrameworkRoot();
|
const frameworkRoot = deps.frameworkRoot ?? resolveFrameworkRoot();
|
||||||
|
|
||||||
@@ -1464,22 +1285,9 @@ export function registerFleetCommand(program: Command, deps: FleetCommandDeps =
|
|||||||
.command(`${action} [agent]`)
|
.command(`${action} [agent]`)
|
||||||
.description(`${action} the fleet holder or one agent`)
|
.description(`${action} the fleet holder or one agent`)
|
||||||
.action(async (agent?: string) => {
|
.action(async (agent?: string) => {
|
||||||
const commandOpts = cmd.opts<{ mosaicHome: string; roster?: string }>();
|
|
||||||
const activePaths = resolveFleetPaths(commandOpts.mosaicHome);
|
|
||||||
const roster = await loadRosterForCommand(cmd);
|
const roster = await loadRosterForCommand(cmd);
|
||||||
if (agent) {
|
if (agent) {
|
||||||
getRosterAgent(roster, agent);
|
getRosterAgent(roster, agent);
|
||||||
// Single-agent restart is guarded too: it can race a full restart that
|
|
||||||
// is tearing the shared holder down.
|
|
||||||
if (action === 'restart') {
|
|
||||||
const guard = await acquireRestartLock(activePaths.mosaicHome, sleepFn);
|
|
||||||
try {
|
|
||||||
await runChecked(runner, buildFleetServiceCommand(action, agent));
|
|
||||||
} finally {
|
|
||||||
await guard.release();
|
|
||||||
}
|
|
||||||
return;
|
|
||||||
}
|
|
||||||
await runChecked(runner, buildFleetServiceCommand(action, agent));
|
await runChecked(runner, buildFleetServiceCommand(action, agent));
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
@@ -1490,21 +1298,6 @@ export function registerFleetCommand(program: Command, deps: FleetCommandDeps =
|
|||||||
);
|
);
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
if (action === 'restart') {
|
|
||||||
// Serialize the holder+agents teardown/relaunch behind the restart lock
|
|
||||||
// so a re-entrant restart waits for clean shutdown before relaunching,
|
|
||||||
// instead of racing a half-torn-down holder into a tight loop.
|
|
||||||
const guard = await acquireRestartLock(activePaths.mosaicHome, sleepFn);
|
|
||||||
try {
|
|
||||||
await runChecked(runner, buildFleetServiceCommand(action));
|
|
||||||
for (const rosterAgent of roster.agents) {
|
|
||||||
await runChecked(runner, buildFleetServiceCommand(action, rosterAgent.name));
|
|
||||||
}
|
|
||||||
} finally {
|
|
||||||
await guard.release();
|
|
||||||
}
|
|
||||||
return;
|
|
||||||
}
|
|
||||||
await runChecked(runner, buildFleetServiceCommand(action));
|
await runChecked(runner, buildFleetServiceCommand(action));
|
||||||
for (const rosterAgent of roster.agents) {
|
for (const rosterAgent of roster.agents) {
|
||||||
await runChecked(runner, buildFleetServiceCommand(action, rosterAgent.name));
|
await runChecked(runner, buildFleetServiceCommand(action, rosterAgent.name));
|
||||||
|
|||||||
161
tools/install.sh
161
tools/install.sh
@@ -16,10 +16,6 @@
|
|||||||
# --framework Install/upgrade framework only (skip npm CLI)
|
# --framework Install/upgrade framework only (skip npm CLI)
|
||||||
# --cli Install/upgrade npm CLI only (skip framework)
|
# --cli Install/upgrade npm CLI only (skip framework)
|
||||||
# --ref <branch> Git ref for framework archive (default: main)
|
# --ref <branch> Git ref for framework archive (default: main)
|
||||||
# --dev Build CLI + gateway FROM SOURCE at --ref instead of the
|
|
||||||
# registry @latest. Zero registry writes — packs local
|
|
||||||
# tarballs and installs them globally. Use to test a branch
|
|
||||||
# end-to-end before cutting a release.
|
|
||||||
# --yes Accept all defaults; headless/non-interactive install
|
# --yes Accept all defaults; headless/non-interactive install
|
||||||
# --no-auto-launch Skip automatic mosaic wizard + gateway install on first install
|
# --no-auto-launch Skip automatic mosaic wizard + gateway install on first install
|
||||||
# --uninstall Reverse the install: remove framework dir, CLI package, and npmrc line
|
# --uninstall Reverse the install: remove framework dir, CLI package, and npmrc line
|
||||||
@@ -31,7 +27,6 @@
|
|||||||
# MOSAIC_PREFIX — npm global prefix (default: ~/.npm-global)
|
# MOSAIC_PREFIX — npm global prefix (default: ~/.npm-global)
|
||||||
# MOSAIC_NO_COLOR — disable colour (set to 1)
|
# MOSAIC_NO_COLOR — disable colour (set to 1)
|
||||||
# MOSAIC_REF — git ref for framework (default: main)
|
# MOSAIC_REF — git ref for framework (default: main)
|
||||||
# MOSAIC_DEV — equivalent to --dev (set to 1)
|
|
||||||
# MOSAIC_ASSUME_YES — equivalent to --yes (set to 1)
|
# MOSAIC_ASSUME_YES — equivalent to --yes (set to 1)
|
||||||
# ──────────────────────────────────────────────────────────────────────────────
|
# ──────────────────────────────────────────────────────────────────────────────
|
||||||
#
|
#
|
||||||
@@ -48,7 +43,6 @@ FLAG_CLI=true
|
|||||||
FLAG_NO_AUTO_LAUNCH=false
|
FLAG_NO_AUTO_LAUNCH=false
|
||||||
FLAG_YES=false
|
FLAG_YES=false
|
||||||
FLAG_UNINSTALL=false
|
FLAG_UNINSTALL=false
|
||||||
FLAG_DEV=false
|
|
||||||
GIT_REF="${MOSAIC_REF:-main}"
|
GIT_REF="${MOSAIC_REF:-main}"
|
||||||
|
|
||||||
# MOSAIC_ASSUME_YES env var acts the same as --yes
|
# MOSAIC_ASSUME_YES env var acts the same as --yes
|
||||||
@@ -56,18 +50,12 @@ if [[ "${MOSAIC_ASSUME_YES:-0}" == "1" ]]; then
|
|||||||
FLAG_YES=true
|
FLAG_YES=true
|
||||||
fi
|
fi
|
||||||
|
|
||||||
# MOSAIC_DEV env var acts the same as --dev
|
|
||||||
if [[ "${MOSAIC_DEV:-0}" == "1" ]]; then
|
|
||||||
FLAG_DEV=true
|
|
||||||
fi
|
|
||||||
|
|
||||||
while [[ $# -gt 0 ]]; do
|
while [[ $# -gt 0 ]]; do
|
||||||
case "$1" in
|
case "$1" in
|
||||||
--check) FLAG_CHECK=true; shift ;;
|
--check) FLAG_CHECK=true; shift ;;
|
||||||
--framework) FLAG_CLI=false; shift ;;
|
--framework) FLAG_CLI=false; shift ;;
|
||||||
--cli) FLAG_FRAMEWORK=false; shift ;;
|
--cli) FLAG_FRAMEWORK=false; shift ;;
|
||||||
--ref) GIT_REF="${2:-main}"; shift 2 ;;
|
--ref) GIT_REF="${2:-main}"; shift 2 ;;
|
||||||
--dev) FLAG_DEV=true; shift ;;
|
|
||||||
--yes|-y) FLAG_YES=true; shift ;;
|
--yes|-y) FLAG_YES=true; shift ;;
|
||||||
--no-auto-launch) FLAG_NO_AUTO_LAUNCH=true; shift ;;
|
--no-auto-launch) FLAG_NO_AUTO_LAUNCH=true; shift ;;
|
||||||
--uninstall) FLAG_UNINSTALL=true; shift ;;
|
--uninstall) FLAG_UNINSTALL=true; shift ;;
|
||||||
@@ -84,17 +72,6 @@ CLI_PKG="${SCOPE}/mosaic"
|
|||||||
REPO_BASE="https://git.mosaicstack.dev/mosaicstack/stack"
|
REPO_BASE="https://git.mosaicstack.dev/mosaicstack/stack"
|
||||||
ARCHIVE_URL="${REPO_BASE}/archive/${GIT_REF}.tar.gz"
|
ARCHIVE_URL="${REPO_BASE}/archive/${GIT_REF}.tar.gz"
|
||||||
|
|
||||||
# In dev (build-from-source) mode the gateway is installed globally from a
|
|
||||||
# locally-built tarball. Tell the wizard / gateway-config stage NOT to overwrite
|
|
||||||
# it with the registry @latest build (honored by gatewayConfigStage).
|
|
||||||
if [[ "$FLAG_DEV" == "true" ]]; then
|
|
||||||
export MOSAIC_GATEWAY_SKIP_NPM_INSTALL=1
|
|
||||||
fi
|
|
||||||
|
|
||||||
# Shared monorepo checkout (populated on demand by ensure_monorepo).
|
|
||||||
WORK_DIR=""
|
|
||||||
EXTRACTED_DIR=""
|
|
||||||
|
|
||||||
# ─── uninstall path ───────────────────────────────────────────────────────────
|
# ─── uninstall path ───────────────────────────────────────────────────────────
|
||||||
# Shell-level uninstall for when the CLI is broken or not available.
|
# Shell-level uninstall for when the CLI is broken or not available.
|
||||||
# Handles: framework directory, npm CLI package, npmrc scope line.
|
# Handles: framework directory, npm CLI package, npmrc scope line.
|
||||||
@@ -262,99 +239,6 @@ framework_version() {
|
|||||||
fi
|
fi
|
||||||
}
|
}
|
||||||
|
|
||||||
# Download + extract the monorepo archive at $GIT_REF exactly once per run.
|
|
||||||
# Sets the script-level EXTRACTED_DIR to the repo root. Reused by both the
|
|
||||||
# framework install (Part 1) and the dev build-from-source path (Part 2).
|
|
||||||
ensure_monorepo() {
|
|
||||||
if [[ -n "$EXTRACTED_DIR" ]] && [[ -d "$EXTRACTED_DIR" ]]; then
|
|
||||||
return 0
|
|
||||||
fi
|
|
||||||
|
|
||||||
require_cmd tar
|
|
||||||
|
|
||||||
WORK_DIR="$(mktemp -d "${TMPDIR:-/tmp}/mosaic-install-XXXXXX")"
|
|
||||||
# shellcheck disable=SC2317
|
|
||||||
cleanup_work() { [[ -n "$WORK_DIR" ]] && rm -rf "$WORK_DIR"; }
|
|
||||||
trap cleanup_work EXIT
|
|
||||||
|
|
||||||
info "Downloading source from ${GIT_REF}…"
|
|
||||||
if command -v curl &>/dev/null; then
|
|
||||||
curl -fsSL "$ARCHIVE_URL" | tar xz -C "$WORK_DIR"
|
|
||||||
elif command -v wget &>/dev/null; then
|
|
||||||
wget -qO- "$ARCHIVE_URL" | tar xz -C "$WORK_DIR"
|
|
||||||
else
|
|
||||||
fail "curl or wget required to download source."
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
|
|
||||||
# Gitea archives extract to <repo-name>/ inside the work dir
|
|
||||||
EXTRACTED_DIR="$(find "$WORK_DIR" -maxdepth 1 -mindepth 1 -type d | head -1)"
|
|
||||||
if [[ -z "$EXTRACTED_DIR" ]] || [[ ! -d "$EXTRACTED_DIR" ]]; then
|
|
||||||
fail "Could not locate extracted source in archive."
|
|
||||||
ls -la "$WORK_DIR" >&2
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
}
|
|
||||||
|
|
||||||
# Build @mosaicstack/mosaic + @mosaicstack/gateway from source and install both
|
|
||||||
# globally from locally-packed tarballs. ZERO registry writes. Workspace deps
|
|
||||||
# (brain/config/db/…) are pulled from the registry at the versions pinned in
|
|
||||||
# each package.json — `pnpm pack` rewrites `workspace:*` to those versions.
|
|
||||||
install_cli_from_source() {
|
|
||||||
local src="$EXTRACTED_DIR"
|
|
||||||
local out_dir="$WORK_DIR/dist-tarballs"
|
|
||||||
mkdir -p "$out_dir"
|
|
||||||
|
|
||||||
# pnpm via corepack (ships with Node >= 16.9; required by Node >= 20 preflight).
|
|
||||||
# Pin to the repo's packageManager version so the build matches CI. Surface
|
|
||||||
# corepack failures so the fresh-machine case gives an actionable error
|
|
||||||
# instead of a bare "command not found".
|
|
||||||
if ! command -v pnpm &>/dev/null; then
|
|
||||||
info "Activating pnpm via corepack…"
|
|
||||||
corepack enable 2>&1 | sed 's/^/ /' || warn "corepack enable failed — pnpm may need manual install."
|
|
||||||
corepack prepare pnpm@10.6.2 --activate 2>&1 | sed 's/^/ /' \
|
|
||||||
|| warn "corepack prepare failed — pnpm may need manual install."
|
|
||||||
fi
|
|
||||||
if ! command -v pnpm &>/dev/null; then
|
|
||||||
fail "pnpm not available after corepack activation."
|
|
||||||
echo " Install pnpm manually (https://pnpm.io/installation) and re-run with --dev."
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
|
|
||||||
info "Installing workspace dependencies (pnpm install)…"
|
|
||||||
( cd "$src" && pnpm install ) 2>&1 | sed 's/^/ /'
|
|
||||||
|
|
||||||
info "Building CLI + gateway from source…"
|
|
||||||
( cd "$src" && pnpm --filter "@mosaicstack/mosaic..." --filter "@mosaicstack/gateway..." run build ) 2>&1 | sed 's/^/ /'
|
|
||||||
|
|
||||||
info "Packing local tarballs…"
|
|
||||||
( cd "$src/packages/mosaic" && pnpm pack --pack-destination "$out_dir" ) 2>&1 | sed 's/^/ /'
|
|
||||||
( cd "$src/apps/gateway" && pnpm pack --pack-destination "$out_dir" ) 2>&1 | sed 's/^/ /'
|
|
||||||
|
|
||||||
local cli_tgz gw_tgz
|
|
||||||
cli_tgz="$(ls -1t "$out_dir"/mosaicstack-mosaic-*.tgz 2>/dev/null | head -1)"
|
|
||||||
gw_tgz="$(ls -1t "$out_dir"/mosaicstack-gateway-*.tgz 2>/dev/null | head -1)"
|
|
||||||
|
|
||||||
if [[ ! -f "$cli_tgz" ]]; then
|
|
||||||
fail "CLI tarball was not produced by pnpm pack."
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
if [[ ! -f "$gw_tgz" ]]; then
|
|
||||||
fail "Gateway tarball was not produced by pnpm pack."
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
|
|
||||||
# Gateway first so it is present globally before the CLI's wizard runs (which
|
|
||||||
# skips its own gateway install via MOSAIC_GATEWAY_SKIP_NPM_INSTALL=1).
|
|
||||||
info "Installing gateway from source tarball (global)…"
|
|
||||||
npm install -g "$gw_tgz" --prefix="$PREFIX" 2>&1 | sed 's/^/ /'
|
|
||||||
|
|
||||||
info "Installing CLI from source tarball (global)…"
|
|
||||||
npm install -g "$cli_tgz" --prefix="$PREFIX" 2>&1 | sed 's/^/ /'
|
|
||||||
|
|
||||||
ok "Installed from source: CLI $(installed_cli_version)"
|
|
||||||
}
|
|
||||||
|
|
||||||
# ─── preflight ────────────────────────────────────────────────────────────────
|
# ─── preflight ────────────────────────────────────────────────────────────────
|
||||||
|
|
||||||
require_cmd node
|
require_cmd node
|
||||||
@@ -398,8 +282,25 @@ if [[ "$FLAG_FRAMEWORK" == "true" ]]; then
|
|||||||
warn "Framework not installed."
|
warn "Framework not installed."
|
||||||
fi
|
fi
|
||||||
else
|
else
|
||||||
# Download repo archive and extract framework (shared with the dev build)
|
# Download repo archive and extract framework
|
||||||
ensure_monorepo
|
require_cmd tar
|
||||||
|
|
||||||
|
WORK_DIR="$(mktemp -d "${TMPDIR:-/tmp}/mosaic-install-XXXXXX")"
|
||||||
|
cleanup_work() { rm -rf "$WORK_DIR"; }
|
||||||
|
trap cleanup_work EXIT
|
||||||
|
|
||||||
|
info "Downloading framework from ${GIT_REF}…"
|
||||||
|
if command -v curl &>/dev/null; then
|
||||||
|
curl -fsSL "$ARCHIVE_URL" | tar xz -C "$WORK_DIR"
|
||||||
|
elif command -v wget &>/dev/null; then
|
||||||
|
wget -qO- "$ARCHIVE_URL" | tar xz -C "$WORK_DIR"
|
||||||
|
else
|
||||||
|
fail "curl or wget required to download framework."
|
||||||
|
exit 1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Gitea archives extract to <repo-name>/ inside the work dir
|
||||||
|
EXTRACTED_DIR="$(find "$WORK_DIR" -maxdepth 1 -mindepth 1 -type d | head -1)"
|
||||||
FRAMEWORK_SRC="$EXTRACTED_DIR/packages/mosaic/framework"
|
FRAMEWORK_SRC="$EXTRACTED_DIR/packages/mosaic/framework"
|
||||||
|
|
||||||
if [[ ! -d "$FRAMEWORK_SRC" ]]; then
|
if [[ ! -d "$FRAMEWORK_SRC" ]]; then
|
||||||
@@ -455,11 +356,7 @@ if [[ "$FLAG_CLI" == "true" ]]; then
|
|||||||
fi
|
fi
|
||||||
|
|
||||||
CURRENT="$(installed_cli_version)"
|
CURRENT="$(installed_cli_version)"
|
||||||
if [[ "$FLAG_DEV" == "true" ]]; then
|
LATEST="$(latest_cli_version)"
|
||||||
LATEST=""
|
|
||||||
else
|
|
||||||
LATEST="$(latest_cli_version)"
|
|
||||||
fi
|
|
||||||
|
|
||||||
if [[ -n "$CURRENT" ]]; then
|
if [[ -n "$CURRENT" ]]; then
|
||||||
dim " Installed: ${CLI_PKG}@${CURRENT}"
|
dim " Installed: ${CLI_PKG}@${CURRENT}"
|
||||||
@@ -467,9 +364,7 @@ if [[ "$FLAG_CLI" == "true" ]]; then
|
|||||||
dim " Installed: (none)"
|
dim " Installed: (none)"
|
||||||
fi
|
fi
|
||||||
|
|
||||||
if [[ "$FLAG_DEV" == "true" ]]; then
|
if [[ -n "$LATEST" ]]; then
|
||||||
dim " Source: ${REPO_BASE} (ref: ${GIT_REF}, build-from-source)"
|
|
||||||
elif [[ -n "$LATEST" ]]; then
|
|
||||||
dim " Latest: ${CLI_PKG}@${LATEST}"
|
dim " Latest: ${CLI_PKG}@${LATEST}"
|
||||||
else
|
else
|
||||||
dim " Latest: (registry unreachable)"
|
dim " Latest: (registry unreachable)"
|
||||||
@@ -477,9 +372,7 @@ if [[ "$FLAG_CLI" == "true" ]]; then
|
|||||||
echo ""
|
echo ""
|
||||||
|
|
||||||
if [[ "$FLAG_CHECK" == "true" ]]; then
|
if [[ "$FLAG_CHECK" == "true" ]]; then
|
||||||
if [[ "$FLAG_DEV" == "true" ]]; then
|
if [[ -z "$LATEST" ]]; then
|
||||||
info "Dev mode: installed version is ${CURRENT:-(none)} (no registry comparison)."
|
|
||||||
elif [[ -z "$LATEST" ]]; then
|
|
||||||
warn "Could not reach registry."
|
warn "Could not reach registry."
|
||||||
elif [[ -z "$CURRENT" ]]; then
|
elif [[ -z "$CURRENT" ]]; then
|
||||||
warn "Not installed."
|
warn "Not installed."
|
||||||
@@ -490,16 +383,6 @@ if [[ "$FLAG_CLI" == "true" ]]; then
|
|||||||
else
|
else
|
||||||
ok "Up to date (or ahead of registry)."
|
ok "Up to date (or ahead of registry)."
|
||||||
fi
|
fi
|
||||||
elif [[ "$FLAG_DEV" == "true" ]]; then
|
|
||||||
info "Dev mode — building CLI + gateway from source at ref ${GIT_REF}…"
|
|
||||||
ensure_monorepo
|
|
||||||
install_cli_from_source
|
|
||||||
|
|
||||||
# PATH check for npm prefix
|
|
||||||
if [[ ":$PATH:" != *":$PREFIX/bin:"* ]]; then
|
|
||||||
warn "$PREFIX/bin is not on your PATH"
|
|
||||||
dim " Add to your shell rc: export PATH=\"$PREFIX/bin:\$PATH\""
|
|
||||||
fi
|
|
||||||
else
|
else
|
||||||
if [[ -z "$LATEST" ]]; then
|
if [[ -z "$LATEST" ]]; then
|
||||||
warn "Could not reach registry at $REGISTRY — skipping npm CLI."
|
warn "Could not reach registry at $REGISTRY — skipping npm CLI."
|
||||||
|
|||||||
Reference in New Issue
Block a user